AMD Beat Nvidia by 53 Points in Five Days. That's a Trade, Not a Thesis.
AMD's 53.2-point five-day blowout over Nvidia is the widest relative move since Nvidia's 1999 IPO. But Friday's 8.75% drop already took a bite out of it, and Nvidia's November 17 earnings will decide whether this was rotation or a head fake.
AMD just posted the biggest five-day win over Nvidia since Nvidia went public, and I'll say the quiet part out loud: that's a positioning event, not a changing of the guard.
The numbers are real, and they're staggering. AMD outperformed Nvidia by 53.2 percentage points over five trading days, according to The Kobeissi Letter. That's the widest gap between the two chipmakers since January 1999, when Nvidia first listed. Twenty-six years of these two names trading against each other, and nothing has looked quite like this.
Then Friday arrived. AMD fell 8.75% in a single session. Grant, one bad day doesn't erase a week like that, but it does tell you something about how much of the move was borrowed.
What a 53-point spread actually means
Relative performance over five sessions measures flows, not fundamentals. A gap that wide means one stock was the crowded long and the other was the funding source. Traders needed cash for something else, or they needed to rotate, and Nvidia has been the ATM for the entire AI trade for two years running.
So the spread tells you where money went. It doesn't tell you why the money should stay.
Here's the part skeptics keep skipping. AMD didn't rally 53 points on a product announcement or a guidance raise. It rallied because it was the less-loved side of a two-stock duopoly. That's a real reason to buy, admittedly, but it's a reason with a short shelf life once the crowd shows up.
The bull case deserves a fair hearing
Now the counterpoint, and it's not weak. Betting on AMD as the second source for AI accelerators has been a losing trade for two years and a winning thesis for five. Proponents will tell you the accelerator market won't stay a one-vendor show forever, that no buyer wants a single supplier holding the pricing power, and that AMD's valuation already prices in years of disappointment.
How much of Nvidia's premium, they ask, is built on the assumption that nothing changes?
That's a fair question. And there's a setup into November 17. Nvidia's earnings print is the single largest event on the semis calendar, and any wobble in guidance sends money hunting for the cheaper expression of the same theme. AMD is that expression. The stock ran because people got early.
My verdict
I'm not entirely convinced the run survives the print. Nvidia has cleared its own bar quarter after quarter, and history suggests otherwise whenever the market decides the leader is finally out of room. Every "Nvidia is done" call since 2023 has been early by at least two quarters, and early is the same as wrong when you're paying for it.
The question worth asking: was this rotation, or was it a squeeze? Rotation keeps going after earnings. A squeeze doesn't.
My read is that the 53-point spread compresses hard once Nvidia reports. That doesn't make AMD a bad company. It makes it a stock that got priced for a story it hasn't told yet. Color me skeptical, but five days of outperformance is a trade, and trades have exits.
What to watch next is simple. Nvidia reports November 17, and the first two sessions after that print will tell you everything. If the spread stays wide, the thesis has legs and I'll eat my words. If it snaps back inside ten points, this was a week-long funding trade dressed up as a rerating.
Time will tell, though. Just not much of it. November 17 is the deadline.
